Legacy software modernisation cost is the question UK businesses put off until something forces it: the server that hosts the old system reaches end of life, the one developer who understands the code retires, or a cybersecurity assessment flags a platform that stopped receiving patches years ago. At that point the finance director asks what fixing it will cost, and the honest answer spans an uncomfortable range — from a five-figure lift onto modern hosting to a six-figure rebuild.
This guide narrows that range with real sterling numbers. It builds on our strategy guide to legacy system modernisation, which explains the rehost, refactor, rebuild and replace routes in depth. Here the focus is money: what each route actually costs in the UK, the day rates behind those prices, and how legacy software modernisation cost behaves over three years rather than in the quote alone.
We price the three routes a decision usually comes down to — rebuild the system from scratch, replatform it onto modern hosting, or replace it with an off-the-shelf product — and then work a full example for a 40-person firm, with every step of the arithmetic shown.
Every figure is a published UK rate, a named public report, or arithmetic you can check. Where a number is an assumption, it is labelled as one. By the end you should be able to put a defensible legacy software modernisation cost estimate in front of a board, and recognise which route a supplier is quietly steering you towards.
Table of contents
- What Drives Legacy Software Modernisation Cost in the UK?
- The Real Cost of Doing Nothing
- Rebuild, Replatform or Replace: The Three Routes Compared
- Legacy Software Modernisation Cost Bands by Route
- UK Day Rates: The Engine Under Every Quote
- Worked Example: Legacy Software Modernisation Cost for a 40-Person Firm
- Hidden Legacy Software Modernisation Cost: What Estimates Miss
- How to Choose: A Decision Framework
- Six Ways to Reduce Legacy Software Modernisation Cost
- Legacy Software Modernisation Cost FAQ
- References
What Drives Legacy Software Modernisation Cost in the UK?
Legacy software modernisation cost is not really a build cost — it is the price of recovering knowledge, moving data, and retiring risk, with some development in the middle. Three drivers set the size of the bill before anyone writes code.
The age tax on every change
Old systems resist change. The UK government’s own State of digital government review found that maintaining legacy systems typically costs three to four times more than maintaining modern equivalents — and the same multiplier applies to every modification made during a modernisation project. A field change that takes an afternoon in a modern codebase can take a week in a 15-year-old one, and that ratio flows straight into the legacy software modernisation cost quote.
Discovery is archaeology, not analysis
On a modern system, discovery means reading documentation and APIs. On a legacy system it means archaeology: working out what the code actually does, which reports the business secretly depends on, and which of the 400 database tables are still written to. Expect discovery to absorb 15–25% of the total legacy software modernisation cost on any system older than ten years — and treat any quote that skips it as a quote for the wrong system.
The knowledge problem prices the risk
The single biggest cost multiplier is whether anyone still understands the system. If the original developers are available, suppliers price the work as engineering. If the system is a black box, they price it as risk — and risk is expensive. Stripe’s developer research found engineers already spend around 42% of their working week dealing with maintenance and bad code; a system nobody understands pushes that ratio, and the price, higher still.
The Real Cost of Doing Nothing
Before comparing routes, put a number on the alternative, because “leave it alone” is not free. The State of digital government review, published in January 2025, found that 28% of central government technology is legacy — up from 26% in 2023 — and that a quarter of surveyed organisations suffered critical service outages during 2024. DSIT’s modelling put the productivity and savings left on the table at roughly £45 billion a year across the public sector.
Private-sector systems age exactly the same way. The takeaway before the chart: every year of deferral, you pay the higher maintenance multiplier and still owe the modernisation bill at the end.
The three quiet bills deferral runs up
The first is the maintenance premium itself — the three-to-four-times multiplier compounding every year. The second is opportunity cost: every integration, report and product idea that gets answered with “the system can’t do that.” The third is exposure: unsupported platforms miss security patches, and the National Audit Office has repeatedly flagged unremediated legacy estates as a primary source of government cyber risk. None of these appear in a budget line, which is exactly why legacy software modernisation cost keeps losing the annual budget argument until an outage wins it.
Rebuild, Replatform or Replace: The Three Routes Compared
The industry catalogue of modernisation strategies runs to seven Rs — AWS’s prescriptive guidance lists rehost, relocate, replatform, refactor, repurchase, retire and retain. For a single business-critical system, the decision almost always collapses to three practical routes, and each behaves differently on cost, risk and payoff.
| Route | What it means | Typical timeline | Delivery risk | What you keep |
|---|---|---|---|---|
| Replatform | Move the existing system onto modern hosting with minimal code change | 1–4 months | Low | All current behaviour — including the limitations |
| Rebuild | Redevelop the system from scratch on a modern stack | 6–18 months | High | Your exact processes, finally without the constraints |
| Replace | Retire the system and adopt an off-the-shelf or SaaS product | 2–6 months | Medium | Your data — but the product’s processes, not yours |
Replatform: pay less, keep the problem contained
Replatforming lifts the system out of its dying environment — the Windows Server 2012 box under the stairs — onto supported cloud infrastructure, changing only what the move forces you to change. It is the cheapest route and the fastest, which is why it dominates when the trigger is an end-of-life deadline. Our guide to Azure migration cost prices this move in detail for UK SMEs. The honest caveat: replatforming resets the hosting clock, not the software clock. The code is as hard to change on Tuesday as it was on Monday.
Rebuild: the highest legacy software modernisation cost, the only full fix
A rebuild redevelops the system on a modern stack, keeping the business logic worth keeping and discarding fifteen years of accumulated workarounds. It is the only route that removes the age tax entirely — and the only one with serious delivery risk. McKinsey and Oxford research across more than 5,400 projects found large IT projects run 45% over budget on average while delivering 56% less value than promised. Budget accordingly, and read our guide on how to rescue a failing software development project before, not after, committing.
Replace: someone else’s software, someone else’s roadmap
If the system does something standard — invoicing, stock, scheduling, HR — a mature product almost certainly does it better than a 2009 bespoke build. Replacement swaps a capital project for a subscription, moves security and maintenance onto the vendor, and shifts the real cost into data migration and process change. The trap is fit: a product that covers 90% of your process leaves the last 10% as manual workarounds or paid customisation, and that gap sets the true legacy software modernisation cost of this route.
Legacy Software Modernisation Cost Bands by Route
The table below puts UK price bands on each route for a single business system. “Small” means one core workflow and a handful of users; “mid-size” means a system central to daily operations for 20–100 staff; “complex” means multiple integrations, regulated data or high transaction volume. Bands assume UK agency delivery — the day-rate section after the table shows how to move them.
| Route | Small system | Mid-size system | Complex or regulated | Recurring after go-live |
|---|---|---|---|---|
| Replatform | £8,000–£25,000 | £25,000–£60,000 | £60,000–£150,000 | Cloud hosting £100–£1,000+ a month, maintenance 15–25% of build a year |
| Rebuild | £30,000–£60,000 | £60,000–£150,000 | £150,000–£400,000+ | Hosting as above, maintenance 15–25% of build a year |
| Replace | £5,000–£20,000 setup | £20,000–£60,000 setup | £60,000–£150,000+ setup | Licences £15–£60 per user a month, forever |
Reading the bands honestly
Two things make these ranges wide. Data is the first: a clean, well-understood database migrates for thousands; a dirty one with two decades of duplicates costs tens of thousands to cleanse whichever route you pick. Integrations are the second: every system the old application talks to needs its connection rebuilt, tested and cut over, and our companion guide to system integration pricing shows each rebuilt connection is a project of its own. A legacy software modernisation cost quote that names neither your data volume nor your integration count is a placeholder, not an estimate.
Where each route’s money actually goes
Replatform money goes on environment work, compatibility fixes and testing — roughly a third each. Rebuild money goes 50–60% on development, 20% on discovery and design, and the remainder on testing and cutover. Replace money splits between implementation partner fees, data migration and training, with the licence bill arriving annually forever after. The recurring column is the one boards skim and should not: over five years, a £30-per-user monthly licence for 50 staff is £90,000 — arithmetic worth doing before calling the replace route cheap. That is 50 × £30 × 12 × 5, and it is why the recurring column belongs in every legacy software modernisation cost comparison.
UK Day Rates: The Engine Under Every Quote
Every legacy software modernisation cost quote decomposes into people multiplied by days. These are the cluster-consistent UK rates we use across all our costing guides, cross-checked against ITJobsWatch contractor medians.
| Role / delivery model | Typical UK day rate | Where it fits |
|---|---|---|
| Mid-level developer (contract) | £450–£650 | Build work on either code route |
| Solution architect | £700–£950 | Discovery, target design, migration plan |
| UK agency (blended) | £550–£750 | Managed delivery with PM and QA included |
| Offshore team (blended) | £150–£300 | Large rebuilds with strong specification |
Why cheap days can cost more on legacy work
Offshore rates look irresistible against a rebuild budget, and for well-specified new builds they can work — our guide to IT outsourcing cost compares the models in full. Legacy work is the hard case for them, because the specification does not exist: it is locked in the old code and in the heads of long-serving staff. Modernisation projects live or die on discovery, and discovery needs people who can sit with your operations team. A common pattern that works: UK-led discovery and architecture, offshore build capacity, UK-led testing and cutover.
Worked Example: Legacy Software Modernisation Cost for a 40-Person Firm
Meet a fictional but realistic 40-person UK distribution firm. Its order-management system was built in 2009 on .NET Framework and SQL Server, runs on an on-premises server due for retirement, and is maintained by one part-time contractor. Three routes, priced with the day rates above, over three years.
Route one: replatform to the cloud
Thirty agency days at £550 covers environment build, compatibility fixes and testing: 30 × £550 = £16,500, plus 20% contingency £3,300 — £19,800 one-off. Cloud hosting at £250 a month is £3,000 a year, and maintenance at 20% of the build is £3,960 a year. Three-year total: £19,800 + £9,000 + £11,880 = £40,680.
Route two: rebuild on a modern stack
Discovery and architecture take a solution architect 15 days at £850 (£12,750); the build takes 110 developer days at £550 (£60,500). Subtotal £73,250, and applying the 20% contingency gives £87,900 one-off. Hosting £3,000 a year; maintenance at 15% of build £13,185 a year. Three-year total: £87,900 + £9,000 + £39,555 = £136,455.
Route three: replace with a SaaS product
A mainstream order-management SaaS at £35 per user a month for 40 users is £1,400 a month — £16,800 a year. One-off costs: implementation partner £12,000, data migration £6,000, training £2,000 — £20,000 in total. Three-year total: £20,000 + £50,400 = £70,400.
The takeaway before the chart: over three years the replatform route costs 30% of the rebuild, and the replacement route sits almost exactly halfway between them.
What the totals hide
The chart is honest about cash and silent about value, and the silence matters in both directions. The replatform total still contains unmodernised code — the age tax continues, so its cheap three years may be borrowed time rather than saved money. The rebuild total buys the only system of the three that fits the firm exactly and costs nothing per user. The replace total keeps growing at £16,800 a year after the chart stops. A fair comparison runs the numbers to year five and prices the old system’s limitations, not just its hosting.
Across rescue projects, the gap between quote and final invoice almost always comes from the same five lines — none of which are development.
Data migration and cleansing
Every route moves data, and legacy data is dirtier than anyone admits: duplicates, orphaned records, fields repurposed years ago for something else. Budget £3,000–£10,000 for a small estate and £10,000–£50,000+ where volumes are large or quality is poor. It is routinely the least-quoted, most-overrun line in the whole legacy software modernisation cost.
Parallel running
For a business-critical system you will run old and new side by side for weeks or months — double hosting, double licences, and staff reconciling between two systems. One to three months of duplicated running cost is a sane allowance inside the legacy software modernisation cost, and cutting it is how businesses convert a safe migration into an outage.
Integration rework
Every connection the old system has — accounting, e-commerce, couriers, reporting — must be rebuilt against the new one and tested end to end. Price each connection as its own small project; a system with five integrations can carry £10,000–£40,000 of integration rework on top of the headline route price.
Training and process change
Replacement changes processes by definition, and rebuilds change screens people have used for fifteen years. Allow £150–£300 per staff member for structured training and floor-walking support, plus a productivity dip in the first month that finance should expect rather than discover.
The old system’s quiet dependencies
Somewhere there is a spreadsheet macro, a scheduled export, or a customer portal quietly reading the old database directly. Discovery should hunt for these explicitly, because each one found in week two is a line item, and each one found at cutover is an incident.
How to Choose: A Decision Framework
The right route falls out of a handful of questions about the system, not from preference. The signals below cover most UK cases; our digital transformation consultants use exactly this triage before any legacy software modernisation cost estimate is written.
| Signal from your system | Points towards | Why |
|---|---|---|
| The workflow is genuinely unusual and wins you business | Rebuild | Off-the-shelf products flatten the edge you charge for |
| The workflow is standard — invoicing, stock, scheduling, HR | Replace | A mature product plus configuration beats bespoke at a fraction of the price |
| The code is sound; only the platform is end-of-life | Replatform | Pay for a move, not a rewrite the code does not need |
| Nobody left understands the code | Replace, or rebuild from behaviour | Replatforming a black box preserves risk at fresh hosting prices |
| A hard deadline is weeks away | Replatform now, decide later | Buy time cheaply, then choose the end-state without a gun to your head |
| This year’s budget is capped | Replatform, phased rebuild after | Spreads the spend across budget years without freezing progress |
The routes combine more often than they compete
The most common successful pattern for mid-sized UK firms is sequential: replatform this year to kill the hosting risk, then rebuild or replace module by module from a position of safety. The worst pattern is the opposite one — a big-bang rebuild launched under deadline pressure against a system nobody documented. That is the exact profile of the projects in the McKinsey overrun data.
Six Ways to Reduce Legacy Software Modernisation Cost
Phase it with the strangler-fig pattern
Instead of one big cutover, new components take over the old system’s functions one at a time, with the old system retiring gradually. Each phase is separately budgetable, separately cancellable, and delivers value before the next begins. It converts one terrifying capital request into a sequence of decisions a board can actually govern, and it is the single most effective structure for controlling legacy software modernisation cost.
Choose the contract shape deliberately
Fixed-price suits the well-specified phases — a replatform, a data migration with agreed volumes. Discovery and rebuild phases price better as time-and-materials with a cap, because a supplier forced to fixed-price the unknown will pad the quote. Our comparison of fixed-price vs time-and-materials contracts shows the overrun data behind that advice.
Modernise the system, not the wish list
Every modernisation attracts a decade of parked feature requests, and each one added to scope inflates the legacy software modernisation cost while delaying the risk reduction you are actually paying for. Ship like-for-like first; enhance from a modern base, where changes cost a quarter as much.
Migrate less data than you think you need
Full history migration is the default request and rarely the right one. Migrating two years of live records and archiving the rest to cheap, queryable storage can cut the data workstream by half. Compliance retention applies to keeping data — not to loading it all into the new system.
Pilot the replace route before committing
For the replace route, a four-week paid pilot with real data on a shortlisted product costs £3,000–£8,000 and regularly saves a six-figure mistake by surfacing the process gaps a demo hides. No reference-checked vendor refuses one.
Time the spend against forced deadlines
End-of-support dates — Windows Server, SQL Server, .NET versions — are published years ahead. Planning eighteen months before the date means competitive quotes and a calm phased plan; planning six weeks before it means paying rush rates for the only route fast enough. The cheapest pound of legacy software modernisation cost is the one spent early.
Legacy Software Modernisation Cost FAQ
What does legacy software modernisation cost for a typical UK SME?
For a single mid-size business system: £25,000–£60,000 to replatform, £60,000–£150,000 to rebuild, or £20,000–£60,000 plus ongoing per-user licences to replace. The system’s data quality and integration count push totals within — and occasionally beyond — those bands.
Is replatforming always the cheapest option?
Upfront, almost always. Over five years, not necessarily: the code keeps its three-to-four-times maintenance multiplier, so a replatformed system that still needs frequent change can quietly push its total legacy software modernisation cost past the price of having rebuilt it. Replatforming is cheapest when the code is stable or when it buys time for a phased rebuild.
How long does a modernisation project take?
Replatforming a single system: one to four months. Replacement with an off-the-shelf product: two to six months including data migration and training. A full rebuild: six to eighteen months. Add parallel running to all three, and distrust any quote whose timeline has no discovery phase at the front.
Can we spread legacy software modernisation cost across budget years?
Yes — that is the strangler-fig pattern’s quiet superpower. Replatform in year one, rebuild or replace modules in years two and three. Each phase stands alone, so a budget freeze pauses progress without stranding a half-finished system.
What happens if we keep deferring it?
You pay the legacy maintenance multiplier every year, accumulate security exposure on unpatched platforms, and eventually make the same decision under emergency conditions at rush-rate prices. The public-sector evidence is blunt: 28% of central government systems are legacy, a quarter of organisations had critical outages in a single year, and the remediation bill did not shrink by waiting.
References
State of digital government review — GOV.UK
A blueprint for modern digital government — GOV.UK
The challenges in implementing digital change — National Audit Office
Delivering large-scale IT projects on time, on budget, and on value — McKinsey
Migration strategies for large-scale migrations — AWS Prescriptive Guidance
Plan your cloud migration — Microsoft Cloud Adoption Framework